Hello Bayron, I’ve been using AI quite heavily for building and refactoring apps, and I keep running into the same pattern:
As long as the project is small → everything feels incredibly productive.
Once the codebase grows → small AI changes sometimes cause unexpected side effects (especially with CSS, layout, or shared components).
Sometimes it’s subtle:
a theme variable changes unexpectedly
a component disappears
a layout shifts without being intended
And debugging that can take more time than the original change.
From what I’ve seen publicly here, Jonsson seems to be one of the few who has implemented a truly larger project using AI extensively.
Jonsson – if you’re reading this – I’d genuinely be interested in seeing your workflow in action.
Not just the result, but:
How you structure prompts
How you control scope
How you prevent side effects
How you review AI-generated changes
I already mentioned in another thread that a short workflow video would be incredibly valuable for the community.
Because I suspect the real differentiator isn’t the tool — but the discipline and process behind using it.
Would love to hear from anyone who has scaled AI beyond small experiments.
Best regards,
Otto